Notes on restriction theory in the primes (Q6594744)

The starting point of this article is the estimate \N\[\N\left(\int_0^1 \left|\sum_{p\le N} u_pe(p\alpha)\right|^{\ell} d\alpha\right)^{1/\ell} \ll_{\ell} \left(\frac{N}{\log N}\sum_{p\le N} |u_p|^2\right)^{1/2} \N\]\Nvalid for \(\ell>2\) proved by \textit{J. Bourgain} [Isr. J. Math. 67, No. 3, 291--311 (1989; Zbl 0692.43005)]. The proof of this inequality was simplified by \textit{B. Green} [Ann. Math. (2) 161, No. 3, 1609--1636 (2005; Zbl 1160.11307)] and \textit{B. Green} and \textit{T. Tao} [J. Théor. Nombres Bordx. 18, No. 1, 147--182 (2006; Zbl 1135.11049)]. An interesting feature of the above result is that it is not valid for \(\ell=2\). The author proves various theorems related to the above inequality. Corollary 1.3 states that \N\[\N\int_0^1 \left|\sum_{p\le N} u_pe(p\alpha)\right|^{2+h} d\alpha\le 7000\frac{\left(1+\frac{3}{2\log N}\right)^h+1/h}{N}\left(\frac{2N}{\log N}\sum_{p\le N} |u_p|^2\right)^{1+h/2} \N\]\Nis valid for all \(N\ge 1000\) and \(h>0\). This provides an optimal transition to \(h=0\). Indeed, by setting \(h=1/\log N\), the author concludes that when \(|u_p|\ll 1\), one has the best possible estimate \N\[\N\int_0^1 \left|\sum_{p\le N} u_pe(p\alpha)\right|^{2+h} d\alpha\ll \sum_{p\le N} |u_p|^2. \N\]\NIn the same vein, the author provides a uniform Hardy-Littlewood majorant property in the sense of \textit{B. Green} and \textit{I. Z. Ruzsa} [Math. Proc. Camb. Philos. Soc. 137, No. 3, 511--517 (2004; Zbl 1066.42009)], namely Theorem 1.4 which states that the inequality \N\[\N\left(\int_0^1 \left|\sum_{p\le N} u_pe(p\alpha)\right|^{\ell} d\alpha\right)^{1/\ell}\le 10^5 \left(\int_0^1 \left|\sum_{p\le N} e(p\alpha)\right|^{\ell}\right)^{\ell} \N\]\Nis valid for all \(N\ge 10^6,~\ell\ge 2\) as soon as \(\sum_{p\le N} |u_p|^2\le \sum_{p\le N} 1.\) The proofs rely on an enveloping sieve stated and proved in the paper as Theorem 4.1, and the precise constants are derived thanks to explicit estimates from the prime number theory from \textit{J. B. Rosser} and \textit{L. Schoenfeld} [Ill. J. Math. 6, 64--94 (1962; Zbl 0122.05001)].